Today, I devoted my time to crafting my inaugural plugin, aided by ChatGPT’s expertise.
This plugin emerged from my desire to seamlessly transfer all my chat interactions into Obsidian while simultaneously purging them from ChatGPT’s history. My aim is to routinely migrate my conversations into Obsidian, ensuring they remain off OpenAI’s servers.
Initially, I contemplated embellishing the plugin with CSS. However, I decided to maintain a markdown-centric approach for broader accessibility and ease of use.
This tool enables the effortless importation of your entire chat history directly from the ZIP file obtained when exporting data from ChatGPT. For more details, please visit the GitHub repository: Obsidian ChatGPT Import Plugin.
One notable feature is the plugin’s ability to bypass empty messages. It’s important to note that occasionally, ChatGPT chats may contain coding errors, particularly within code blocks. Unfortunately, these cannot be rectified by the plugin, which might result in occasional markdown discrepancies. These are dependent on the original chat source.
I eagerly welcome your feedback and suggestions. I am also seeking assistance in integrating a folder suggestion feature into the settings. Currently, users must manually input the path, and any help to improve this aspect would be greatly appreciated.
so there is no API to get the export from gpt I assume? too bad
and one more stupid question. since openai already had you chat, do you think it makes any sense to export/delete them? except ofc to have a local copy. I mean, they are still on their servers, even after you deleted them. or is there a statement saying otherwise? GDPR?
Damn I totally forget to come back here sometimes. Sorry for the delay. I didn’t dig into API for ChatGPT. But given it’s ChatGPT and not GPT API, my guess is: No.
That said, I should take some time to get back to improving the plugin. There are most certainly improvments that would benefit this. And ChatGPT evolving I also have to ensure it still works.
When I started, ChatGPT vision was not a thing. I have to check if/how this could eventually be handled. I just need time and energy.
For me the idea of importing conversations in Obsidian make sense to optimize notes, dataflow. I have extensive conversations with ChatGPT about many topics about health, business creation, coding, and most of the conversations I have contains a lot of valuable information I want to be able to index and find easily from within my second brain which is Obsidian (well, I’m a newbee… Lots of work to make it my second brain).
This beta release introduces several key improvements and new features to enhance the ChatGPT Import experience in Obsidian.
New Features:
Import log file with statistics and detailed results of the import process
Option to add date prefix to folder names
ZIP file validation to ensure integrity of imported data
Incremental naming for identical file names to prevent overwriting
Processed ZIP file catalog to prevent unnecessary reprocessing
Improvements:
Fixed issue where note updates were deleting previous messages
Corrected update date not refreshing in existing notes
Changed date formatting to use localized format
Refactored code for better readability and maintainability
Updated documentation for clearer user guidance
Known Issues:
Testing has been conducted on Linux and Mac; Windows experiences may vary
File name contains extend characters. This has to be sanitized
Some notes with identical names might be updated rather than incremented with a suffix
Note filenames may include special characters
Upcoming Improvements (To-Do):
Enhance counter functionality
Further improve the import log
Add option to enable/disable target folder selection for each ZIP import
Modify log filename to include ZIP file date or reference (specifics TBD)
Sanitize filenames to remove special characters
Add support for other chat platforms
We appreciate your feedback as we continue to refine and improve the plugin. Please report any issues or suggestions through our GitHub issues page. Windows users are especially encouraged to provide feedback on their experience with the plugin.
I’m excited to announce the release of version 1.0.2 of the Nexus AI Chat Importer plugin. This update brings several improvements and new features to enhance your experience with importing and managing AI chat conversations in Obsidian.
What’s New:
• Import conversations from ChatGPT export files into Obsidian
• Automatically organize conversations by date
• Create individual Markdown files for each conversation
• Open the original ChatGPT conversation directly from Obsidian with a single click
• Improved import reports with more accurate and detailed information
Enhancements:
• Refactored and optimized code for better performance
• Improved file handling and import logic
• Enhanced error handling and logging
Bug Fixes:
• Fixed issues with import report accuracy
• Resolved problems with dialog functionality
• Improved compatibility with Obsidian’s Reading view
Important Notes:
• The plugin has been renamed to “Nexus AI Chat Importer”
• New metadata parameters have been added for enhanced functionality
• It’s recommended to re-import your conversations for optimal performance
Thank you for your continued support and feedback. If you encounter any issues or have suggestions for future improvements, please don’t hesitate to reach out on GitHub Issues or in this thread.
Hi there. I wanted to import all my ChatGPT chats and went through the process but it skipped more than it imported. From 444 conversations, it only created 192 notes and skipped 251 of them. There is no logical reason for skipping as every conversation contains content.
Any suggestions would be appreciated.
At 1st zip, it should create 100% of notes, unless they are reason to ignore some like empty conversations or invalid.
At 2nd zip, it will compare with already existing conversations, and if there are redundancies, it will check if there are new messages. If conversations are identical, they will be skipped.
i.e. my 1st zip contains 494 conversations, my second 495.
1st pass, it creates 494.
2nd pass, it creates 1, update 1, and skipp 493.
Let me know the details, or share the report if it’s ok for you. You can share in private if you prefer for privacy (or ask chatGPT to scramble the report to hide private data )
What’s New:
Enhancements:
Bug Fixes:
Important Notes:
Links: